Imagination Reduces False Memories for Everyday Action Sentences: Evidence From Pragmatic Inferences

Abstract

Human memory can be unreliable, and when reading a sentence with a pragmatic implication, such as “the karate champion hit the cinder block,” people often falsely remember that the karate champion “broke” the cinder block. Yet, research has shown that encoding instructions affect the false memories we form. On the one hand, instructing participants to imagine themselves manipulating the to-be-recalled items increase false memories (imagination inflation effect). But on the other hand, instructions to imagine have reduced false memories in the DRM paradigm (imagination facilitation effect). Here, we explored the effect of imaginal encoding with pragmatic inferences, a way to study false memories for information about everyday actions. Across two experiments, we manipulated imaginal encoding through the instructions given to participants and the after-item filler task (none vs. math operations). In Experiment 1, participants were either assigned to the encoding condition of imagine+no filler; pay attention+math; or memorize+math. In Experiment 2, the encoding instructions (imagine vs. memorize) and the filler task (none vs. math) were compared across four separate conditions. Results from the two experiments showed that imagination instructions lead to better memory, by showing a higher proportion of correct responses and better performance in a memory benefit index. Similarly, a significant reduction of false memories was observed across both experiments, even though a complementary Bayesian analysis only supported this conclusion for Experiment 1. The findings show that imaginal encoding improves memory, suggesting the engagement of a distinctiveness heuristic and source-monitoring process.

Can we learn from errors? Retrieval facilitates the correction of false memories for pragmatic inferences

“…The list of experimental sentences used and the corresponding coding criteria for correct responses and pragmatic inference errors can be found in the supplementary material. Like in previous protocols [50], responses at the intermediate cued-recall test for those in the retrieval condition, and in the final cued-recall test for all participants were recoded following an adaptation of the standard coding procedure from Brewer [46], resulting in four response types: i) correct responses, which corresponded to those answers matching the original sentence or synonyms maintaining the original meaning of the sentence (i.e., "put his lips towards", "approached" for the sentence "The charming prince gently put his lips towards Snow White's cheek"); ii) pragmatic inference responses, which included responses that matched the expected pragmatic inferences or their synonyms (i.e., "kissed"), meeting the butnot test (i.e., "The charming prince gently put his lips towards Snow White's cheek, but did not kiss her"); iii) intrusions, corresponding to other alternative answers (i.e., "touched", "smelled","looked"); and iv) omissions, for responses left blank. The proportion of each response type was calculated, resulting in four dependent variables.…”

“…After the presentation of each sentence, participants had 5 seconds to perform an easy arithmetic operation (e.g., 23–5 = ?) to prevent retrieval, as used in previous protocols [ 50 ]. Before the start of the task, participants completed a practice trial with five sentences (without pragmatic implications).…”

“…At the end of the study, participants responded to a brief self-report questionnaire about their attention and environmental conditions during the study (similar to Experiment 2 in [ 50 ]). Participants were asked to rate their level of attention, the quality of their data, and had a blank space to provide any comments in relation to their performance.…”

“…Participants. The sample size was determined according to sample resources availability on a priori grounds-based on previous studies with similar experimental protocols manipulating error correction [44] or using pragmatic inferences as experimental material [50]. 120 university students from the University of Lisbon participated in the experiment (M age = 21.05 ± 6.24; 100 female) and were rewarded with 10€ vouchers for their participation.…”

Psychological Applications and Trends 2022

“…Overall, findings have shown that imagining actions makes them as vivid and real as their actual realization (Lyle & Johnson, 2006;Mitchell & Johnson, 2009). In contrast, few studies have shown a reduction of false memories for imagined action sentences (Maraver et al, 2021).…”